I'm Bihan, and I lead product at Decagon, a Series B startup building AI agents for customer service and support use cases.

We're creating a better, more human-like customer support experience with AI agents to help users get the support they need efficiently.

The development of AI agents involves creating systems that can autonomously handle tasks and communicate issues back to users for resolution.

Customer support is an interesting space for AI agents because it involves a lot of unstructured information that can be effectively utilized by AI models.

The market for AI customer support agents is crowded, but many existing solutions are inadequate for complex workflows, leaving room for more sophisticated entrants.

During the deployment process, we work closely with customers to define how their AI agent should respond, including tone and response structure, to meet their specific needs and preferences.

Cisco employs a 'supervisor approach' to decompose complex customer queries across specialized AI agents, improving response quality.

AI is redefining customer support, becoming the most important aspect of our business.

The admin dashboard is crucial for customers to build and configure their AI agents, allowing them to test and roll out the agent safely while monitoring performance and user interactions.